AWS Pricing Calculator Skill

Generate a shareable AWS Pricing Calculator URL from any architecture description.

Trigger Terms

Workflow

Follow these 6 steps in order:

Step 1: Extract Services

Read the architecture document, blog post, or user description. Identify:

• Each AWS service used
• Instance types, storage sizes, throughput, node counts
• Region (default: us-east-1)
• Grouping (environments, tiers, components)

Step 2: Discover Schemas

For each service, run the discovery script to get the current schema:

python ~/.claude/skills/aws-pricing-calculator/scripts/calc_discover.py <serviceCode1> <serviceCode2> ...

This fetches the live service definition from CloudFront and extracts all configurable calculationComponents with their IDs, types, options, and defaults.

Check references/service_formats.md first — if the service has a proven format there, use it directly instead of discovering from scratch.

Step 3: Look Up Pricing

Use the AWS Pricing API MCP tool (get_pricing) to get current prices:

get_pricing(serviceCode, region, filters)

Calculate monthly costs: unit_price * quantity * hours_per_month (730 hrs).

Step 4: Build Estimate

Create a JSON spec file defining groups and services, then build:

python ~/.claude/skills/aws-pricing-calculator/scripts/calc_build.py spec.json -o estimate.json

Or use the Python API directly to build the estimate dict programmatically.

Step 5: Save to API

Upload the estimate to get a shareable URL:

python ~/.claude/skills/aws-pricing-calculator/scripts/calc_save.py estimate.json

Step 6: Return Results

Provide the user with:

1. The calculator URL
2. A cost summary table (service, monthly, annual)
3. Total monthly and annual costs

Tools

Critical Rules

1. calculationComponents must NOT be empty — populate all required fields or the calculator shows $0.00
2. serviceCost.monthly must be pre-calculated — the calculator displays this value as-is on load; it does not recalculate
3. estimateFor must match the template ID from the service definition (e.g., "template" for EC2, "appStream2" for AppStream)
4. version must match the current service definition version — always fetch fresh via calc_discover.py
5. Use curl subprocess, not Python requests — Python 3.14 has SSL issues with CloudFront CDN endpoints
6. Use AWS Pricing API MCP (get_pricing) for real price lookups — never hardcode prices
7. Check references/service_formats.md for known working formats before building from scratch

References

Quick Start Examples

From a blog post URL

User: "Create an AWS pricing calculator for this blog post: https://aws.amazon.com/blogs/big-data/..."

1. Fetch and read the blog post
2. Extract architecture: services, instance types, storage, throughput
3. Run through Steps 2-6

From a solution document

User: "Cost this architecture: 3 EC2 m5.xlarge Windows, 1 RDS Oracle db.m5.xlarge, EBS io1 1400GB"

1. Parse the service specs from the description
2. Discover schemas for ec2Enhancement, amazonRdsForOracle, amazonElasticBlockStore
3. Look up pricing, build estimate, save, return URL

From a CSV/spreadsheet

User: "Build a calculator estimate from this CSV with our environment sizing"

1. Read the CSV to extract services per environment
2. Map to service codes and calculationComponents
3. Build grouped estimate, save, return URL